Yazar "Dalkilic, Ahmet Selim" seçeneğine göre listele
Listeleniyor 1 - 3 / 3
Sayfa Başına Sonuç
Sıralama seçenekleri
Öğe Discharging performance prediction of experimentally tested sorption heat storage materials with machine learning method(Elsevier, 2022) Çolak, Andaç Batur; Aydin, Devrim; Al-Ghosini, Abdullah; Dalkilic, Ahmet SelimIn this study, the usability of the machine learning method in predicting the discharge performance of experi- mentally tested sorption heat storage materials was investigated. Experimental data was obtained from a lab scale fixed-bed thermochemical heat storage unit. 9 candidate composites were tested under different inlet conditions. Based on the experimental data, moisture sorption rates, heat output, exergy output and energy storage densities were determined. For the 6 cycles testing, highest average heat and exergy output were ob- tained with vermiculite/LiCl composite with the values of 0.83 kW and 0.013 kW, respectively. On the other hand, P-CaCl2 was found as the most durable material in terms of energy storage density (296 ? 209 kWh/m3). A multilayer perceptron artificial neural network was established to evaluate measured data and its prediction performance was extensively studied. In the model 54 experimental data sets were utilized, consisting of 6 cycles testing of 9 different composite sorbents. Levenberg-Marquardt algorithm was benefited as the training one in the artificial neural network model established and the Tan-Sig and Purelin functions were selected as the transfer one in the multilayer neural network with 7 neurons in the hidden layer. According to the mathematical defi- nition of the discussed statistical metrics, experimental data were used to compare them to the predicted output in order to verify the reliability of the proposed ANN model; and the analysis of the model was performed by examining the coefficient of determination, mean squared error, and deviation values, which were assumed as performance parameters, in detail. The deviation rate between the prediction values acquired from the artificial neural network and the practical data was determined as less than ±5 %. The acquired findings showed that artificial neural networks, which is one of the common machine learning algorithms, is a preferable method that can be employed to estimate the discharge performance of sorption heat storage materials.Öğe Prediction of heat transfer characteristics in a microchannel with vortex generators by machine learning(Walter de Gruyter, 2022) Gönül, Alişan; Çolak, Andaç Batur; Kayaci, Nurullah; Okbaz, Abdulkerim; Dalkilic, Ahmet SelimBecause of the prompt improvements in Micro-Electro-Mechanical Systems, thermal management necessities have altered paying attention to the compactness and high energy consumption of actual electronic devices in industry. In this study, 625 data sets obtained numerically according to the change of five different geometric parameters and Reynolds numbers for delta winglet type vortex generator pairs placed in a microchannel were utilized. Four dissimilar artificial neural network models were established to predict the heat transfer characteristics in a microchannel with innovatively oriented vortex generators in the literature. Friction factor, Nusselt number, and performance evaluation criteria were considered to explore the heat transfer characteristics. Different neuron numbers were determined in the hidden layer of each of the models in which the Levethenberg–Marquardt training algorithm was benefited as the training algorithm. The predicted values were checked against the target data and empirical correlations. The coefficient of determination values calculated for each machine learning model were found to be above 0.99. According to obtained results, the designed artificial neural networks can provide high prediction performance for each data set and have higher prediction accuracy compared to empirical correlations. All data predicted by machine learning models were collected within the range of ±3% deviation bands, whereas the majority of the estimated data by empirical correlations dispersed within ±20% ones. For that reason, a full evaluation of the estimation performance of artificial neural networks versus empirical correlations data is enabled to fill a gap in the literature as one of the uncommon works.Öğe Research on the influence of convector factors on a panel radiator’s heat output and total weight with a machine learning algorithm(Springer, 2023) Calisir, Tamer; Çolak, Andaç Batur; Aydin, Devrim; Dalkilic, Ahmet Selim; Baskaya, SenolIn the current work, the impacts of convector factors of a panel radiator regarding heat output and total weight have been investigated using a machine learning algorithm. An artificial neural network model, widely evaluated by machine learning algorithms, has been created to determine the heat output and total weight values of panel radiators. There are 10 neurons in the hidden layer of the machine learning model, which was trained using 111 numerically obtained data sets. A comprehensive numerical investigation has been done for dissimilar geometrical dimensions of convectors evaluated in panel radiators and validated with experimental results. Afterward, the Levenberg–Marquardt structure has been employed as the training one in the multilayer perceptron network structure. The heat output and total weight outcomes acquired from the artificial neural network have been contrasted with the computational data and the compatibility of the data has been examined comprehensively. Furthermore, various performance parameters have also been determined and the estimation performance of the neural network has been examined thoroughly. The mean deviation values for the thermal power and weight values gained from the network structure have been determined as 0.04 and 0.004%, in turn, and the R-value has been obtained as 0.99999. The investigation outcomes indicated that the proposed neural network can forecast the heat output and total weight values of the panel radiator with very high accuracy
